Modern Impacts on an Ancient Landscape, the Piedmont Plain in Southwest Turkmenistan

Modern impacts on an ancient landscape, the piedmont plain in southwest Turkmenistan

Berking, J., Beckers, B., Reimann, T., Pollock, S., & Bernbeck, R.

This article is made publically available in the institutional repository of Wageningen University and Research, under article 25fa of the Dutch Copyright Act, also known as the Amendment Taverne.

Article 25fa states that the author of a short scientific work funded either wholly or partially by Dutch public funds is entitled to make that work publicly available for no consideration following a reasonable period of time after the work was first published, provided that clear reference is made to the source of the first publication of the work.

For questions regarding the public availability of this article, please contact [email protected].

Please cite this publication as follows:

Berking, J., Beckers, B., Reimann, T., Pollock, S., & Bernbeck, R. (2017). Modern impacts on an ancient landscape, the piedmont plain in southwest Turkmenistan. Wiley Interdisciplinary Reviews: Water, 4(2), [e1202]. https://doi.org/10.1002/wat2.1202 Focus Article Modern impacts on an ancient landscape, the piedmont plain in southwest Turkmenistan Jonas Berking,1* Brian Beckers,1 Tony Reimann,2 Susan Pollock3 and Reinhard Bernbeck3

The piedmont plain in southwestern Turkmenistan has experienced a millennia- long settlement history despite prevailing arid climates. One of the prerequisites for the various agricultural efforts was irrigation. Most of the water used for irrigation measures came from the adjacent Kopet Dag mountain chain. This situation changed with the introduction of the Karakum canal in the middle of the 20th century. The present study evaluates the rich irrigation history of the piedmont plain by investigating two small catchments that drain the eastern ranges of the Kopet Dag. Within their catchments, geomorphological and hydrological analyses were conducted. We present several Optically Stimulated Lumines- cence and 14-C dating results that add to the understanding of the landscape history from the Pleistocene until modern ages. Moreover, modern climatological and hydrological data were analyzed that show a remarkable drop in runoff from the Kopet Dag since the 1960s. © 2017 Wiley Periodicals, Inc.

How to cite this article: WIREs Water 2017, 4:e1202. doi: 10.1002/wat2.1202

INTRODUCTION possible through irrigation measures. The area is among the earliest ones in Central Asia where agricul- – he piedmont plain in southwest Turkmenistan tural activities, and hence irrigation, took place.1 3 Tis an arid and nowadays remote area. How- As in many arid environments, agricultural suc- ever, numerous settlement remains in the shape of cess was mainly controlled by the highly variable small mounds (locally called depes) indicate that spatial and temporal availability of water. Despite the area flourished in history and prehistory. For subsistence, these settlements relied on agriculture, these challenges, vast drylands have been culturally raising of animals, and hunting, the first being only highly dynamic areas in which the inhabitants have developed multiple ways to cope with these uncer- 4,5 *Correspondence to: [email protected] tainties. The art of managing drylands and mitigat- 1 ing their risks is of significant interest to many Department of Earth Sciences, Physical Geography, Freie Univer- – sität Berlin, Berlin, Germany studies and applications.6 9 2Netherlands Centre for Luminescence Dating, Soil Geography The special character of the area in focus is its and Landscape group, Wageningen University, Wageningen, The very long and quasi-continuous settlement history Netherlands coupled with its irrigation strategies. Moreover, it is 3Institute for Ancient Near Eastern Archaeology, Freie Universität Berlin, Berlin, Germany a good example of the numerous central Asian riverine or oasis landscapes which have been nuclei of Conflict of interest: The authors have declared no conflicts of inter- 10 est for this article. human settlement activities throughout history. This changed during the last century when the fi [Correction added on 16 February 2017, after rst online publica- Soviet regime built the Karakum Canal in the tion: Appendices have been added at the end of the article as these were initially published as Supporting Information.] region, which detaches regional irrigation measures

FIGURE 1 | General climatic setting of Turkmenistan and the surrounding countries. Gray layers depict the annual mean precipitation (data from 1950 to 2000, worldclim.org database). Indicated in italic bold letters are the Koeppen–Geiger climate zones. The red box delineates the area of interest, see also Figure 2. from the dependency on natural water availability the region. The hydrology of these two rivers, the and supply. Meana [Meana (loc. Miana) = name of settlement The study area is characterized by low precipi- and river] and Chacha [Chacha (loc. Çaaça or Chaa- tation and a high amount of insolation (Figure 1). cha) = name of settlement and river], are assumed to Even without any human impact, these dryland eco- be representative for the problems stated above. systems with their sparse or initial vegetation and soil During the fieldwork and informal interviews cover are vulnerable to the impacts of climate with locals in the region in 2011 and 2012, we were change, earthquakes, or floods.9 told that these rivers became ever dryer and their The main question tackled in this study is: how runoff behavior changed within the last decades. In irrigation and management of water supply changed the living memories of local residents, they were once through the long history of the study region, including perennially flowing streams, which at the time con- the major shift that occurred with the construction of tributed the majority of water available for irrigation the Karakum canal. In this sense, the scope of the and drinking purposes. Such a perennial, rather than present study is to combine and discuss data from the current ephemeral, river regime would offer an (1) sediment chronologies, (2) archeological investiga- explanation for the long history of irrigation in the tions, and (3) modern climatic and hydrological data. area. However, in the modern situation the rivers are Much of the analysis and results presented dry for much of the year and deep wells (more than focus on the discharge history of two small rivers in 15 m) no longer reach the groundwater.

60°20′E 60°40′E

Physical

Loess hills Alluvial fan (active)

37°N Alluvial fan (innactive) 37°N Catchment Main streams

Elevation (m asl)

0–250 250–500 500–750 750–1000 >1 000

Other N ′ N Sampling site ′

36°50 Ancient settlements 36°50 Modern settlements

Border Street

Kilometers 0 5 10

60°20′E 60°40′E

FIGURE 2 | Study site map with the main landscape features.

Historical Background Kopet Dag.12,13 In some of the excavated settlements, Turkmenistan´s agricultural sector depends on irriga- cereals such as barley or einkorn have been found, tion, as is the case for most of central Asia. Conse- pointing to the earliest agricultural endeavors (Ref 14; quently, access to water plays a major role in its Miller in Ref 15). The earliest approach to irrigation political, economic and social history. The longue may have combined naturally favorable locations durée history of irrigation in the region spans more such as the floodplains of the rivers that emerge from than 8000 years, from its beginnings in the Neolithic the Kopet Dag, with simple ditch and canal construc- 1,3,16–18 through the Tsarist and later Soviet Union rule, tions to raise the soil water content. before the modern state of Turkmenistan emerged in Settlement activity during the Aeneolithic and 1991.11 later Bronze Age is evidenced by sites on the pied- Since more than 90% of Turkmenistan is virtu- mont plain that most probably utilized the same ally uninhabitable desert, major settlements are— schemes of floodwater irrigation, but with increasing today as in the past—bound to locations of water scales. From the fourth millennium onward, these surplus. These locations are the strips along the major settlements grew significantly in size, and large urban rivers and the piedmont region at the Kopet Dag centers such as Altyn Depe and Gonur Depe emerged footslopes, the latter of which is the focus here. The (for the location of Altyn Depe, see Figure 2). Later, earliest examples of settlements date to the Neolithic from the first millennium BCE to the first millennium to early Aeneolithic periods of the late seventh CE, Iron age, Achaemenid, Seleucid, Parthian, and through fifth millennia BCE, the remains of which are Sasanian settlement systems emerged. At least since dispersed over much of the piedmont region of the Sasanian time, the ever growing importance of the

Silk Road, which passed through the plains north of The two rivers in the focus of this study, the the Kopet Dag, must have boosted the local Meana and Chacha, are small dryland rivers, repre- economy.19 sentative of many small streams that emerge from the The last two centuries brought a distinct change higher Kopet Dag Mountains. They show a sharp in irrigation schemes, because the distant and centra- decrease in their velocity when flowing through the lized governments of the Russian Tsarist and later piedmont plain and later into in the Karakum Desert Soviet rule introduced and installed modes and where they desiccate.25 The two streams are each means of water management and large-scale irriga- characterized by a set of conically shaped active and tion plus agricultural planning. This culminated in inactive alluvial fans. Today they are ephemeral and the construction of the longest canal in the world, only carry water during and after the rainy season or the Karakum canal, between 1954 and 1988, which snow melt.1 Their recent wadi beds are several annually redirects approximately 13 km3 of water meters deep and up to 20 m wide. Framed by the from the Amu Darya river and transports the water alluvial fans is an area that is not directly shaped by through Turkmenistan along the fringe of the Kara- the periodic wadi runoff, but rather by surface runoff kum Desert, with a length of approximately after rainfall events (cf. Ref 25). Multiple archeologi- 1375 km.20 Two consequences of the construction of cal sites have been identified there, among them this canal were: (1) the availability of vast newly cul- Monjukli Depe (MD) with a stratigraphy spanning tivable areas (about 1 million ha) now detached from the late seventh to mid-fifth millennia BCE, and the their local water dependency, and (2) the slow desic- much larger, third millennium BCE Bronze Age city cation of the lower Amu Darya reaches, manifested of Altyn Depe (for early periods: Refs 15,26–28; in the shrinking and desiccation of the Aral Sea.11,20 Altyn Depe: Ref 29; see also Ref 30). Like other former Soviet states, Turkmenistan’s land The piedmont plain is characterized by sparse and water management experienced problems since vegetation and high Eolian activity. This high activity its independence in 1991. Especially the loss of the is due to prevailing winds from northern directions, former management of irrigation systems, mainly for which transport mainly silt-sized particles from the the cotton industry, and the high cost of maintaining inner deserts to the mountainous foreland, where the Karakum canal have led to a severe deterioration they precipitate as loess.31 Soils are only weakly of the water structures.21 developed, mostly as initial AC-soils (Syrosems), and no paleo-horizons could be identified. Salty crusts on abandoned fields are visible and solonchaks, salty Regional Setting soils, emerge where standing water occurs over a The area of interest is the transitional zone between longer period.32 the Kopet Dag Mountains and the Karakum Desert The climate of the area is twofold with dry cli- in southeastern Turkmenistan. Fieldwork was under- mates and a high variability in the lowlands and wet- taken in the summers of 2011 and 2012 in the area ter conditions in the higher regions, which is mainly surrounding the modern villages of Meana and Cha- due to the continentality and topography of the cha (Figure 2). area.33 The region is classified as a dry steppe The area is best described as a piedmont zone (Koeppen–Geiger Classification: BSk) with very hot that interlinks the dominating erosional processes in and dry summers, cold and mostly dry winters and a the highlands with the depositional processes in the short rainy season in early spring. lowlands. The resulting landforms can accordingly This distinct annual divide is due to atmos- be described as erosive (pediments, glacis) or deposi- pheric patterns that are either influenced by offshoots tional (alluvial fans).22 of the stable Siberian anticyclone, leading to very The upper Kopet Dag consist primarily of Cre- cold and dry conditions, or interrupted by the intru- taceous and Paleogene sandstone and limestone, sion of warmer air masses mostly from the north- leading to erosional material composed of sandy, west, bringing relatively warm and humid air. These loamy or marl sediments.23 The Kopet Dag Moun- conditions are the main factors influencing the availa- tains are tectonically active, which is also expressed bility of moisture in the upper elevations, which often in their linear mountain ranges (strip faulting) along precipitate as snow. During the summer period, the with recent earthquake activity; the 1948 earthquake radiation from the bare surfaces is the main factor of 7.2 on the Richter scale killed more than 100,000 that leads to the cessation of nearly all cyclonic activ- people in the capital Ashgabat.24 The lower part of ity and lowers the humidity level (cf. Ref 31; further the Kopet Dag is covered by a thick loess layer which climate analyses are presented in the Results and Dis- gradually merges with the piedmont plain (Figure 3). cussion section.)

FIGURE 3 | Three-dimensional visualization of the study site. Blue lines indicated water-bearing channels on the 1:50,000 topographic maps (reference see below).

The area is scattered with agricultural fields Syrosem topsoils, supporting sparse vegetation. The and small irrigation canals, many of them probably sediment properties were studied macroscopically for cotton growth. Since direct precipitation in the and in situ. piedmont is not sufficient to sustain rain-fed agricul- Russian topographic maps [Turkmenskaja SSR, ture, irrigation is mandatory. It appears that since 1:50,000, Generalńyj Stab, Moskva (1954–1958, the construction of the Karakum canal, all irrigation renewed 1973), map sheets: J-41-109-B, J-41-109-G, water is taken from it. J-41-110-A, and J-41-110-V] turned out to be very helpful by (1) providing detailed information on archeological remains and stream patterns and METHODS AND MATERIALS (2) representing the setting before 1950, when the fi Special interest during field surveys was given to landscape was probably much less modi ed than locations with the potential to provide insights into today (cf. Ref 34). the river and sediment history of the region. The investigation of the river bed sediments was restricted to the lower parts of the catchments; the upper por- tions are located in Iran and crossing the border was Hydrological, Climatological, and Digital impossible. All landscape archives, sample points, Data Treatment and areas of interest were recorded using a standard Terrain and relief information was derived from a GPS, described in the field and later combined in Digital Elevation Model (SRTM 1, acquisition 2000, Geographic Information Systems (GIS). published 2014, http://earthexplorer.usgs.gov/). The sediment profiles analyzed show rather Hydrological properties, such as catchment delinea- homogenous sequences of either (1) fluvial sediment tion, stream network generation, and vertical stream (sand and gravel) or (2) Eolian or fluvially redepos- distance were calculated in Saga-GIS on the basis of ited silt or loess facies. No distinct layering or even the SRTM (for Saga-GIS, see: http://www.saga- (paleo-)soil horizons were found, apart from initial gis.org).

Monthly river discharge (m3/second) was during nighttime. Additional material was sampled obtained from the UNESCO water database which for gamma spectrometry. In the laboratory, activity provides quasi-continuous data of the Meana and concentrations of 40K and several nuclides from the Chaacha rivers from 1936 to 1974 and more scat- uranium and thorium decay chains were measured tered data from 1981 to 1985. Since the gauging sta- using high-resolution gamma-ray spectrometry. tions no longer exist, we approximated their location OSL dating for two samples (LUM 1 + 4) from (data: http://webworld.unesco.org/water/ihp/db/ the 2011 campaign was carried out by the Lumines- shiklomanov/). cence Dating Labor OSL Laboratory (OH, USA; Daily precipitation data were obtained from http://www.laberosl.com/). The laboratory utilized Global Historical Climatology Network (GHCN) bulk (MAR-protocol) values from a Daybreak and from the International Research Institute for Cli- Reader. Due to this older methodology, the results mate and Society (IRIDL; downloaded 2015 from should be treated with caution. The age results are http://iridl.ldeo.columbia.edu). Gridded precipitation shown in Table 2. data for annual means 1950–2000 were obtained OSL dating for the other four samples (Turk- from the Worldclim database and the DEKLIM OSL 1, 2 and Turk-OSL 5, 6) from the 2012 field VASClim project.35,36 The next available weather campaign was carried out by the Netherlands Centre stations are Kaka and Hawdan (Figure 1). Quasi- for Luminescence dating (NCL; Wageningen, The continuous data are available for Kaka 1936–1965 Netherlands). For OSL dating, two quantities are and 1977–1991. For Hawdan, it is 1936–1961. determined: the received accumulated dose during Because most of the variance of monthly river burial (palaeodose) and the environmental dose rate discharges can be explained by the precipitation in of the sample surrounding. For dose rate determina- their upper catchments, the available monthly cli- tion, the activity concentrations of 40K and typical matic datasets for the upper regions were investigated nuclides from uranium as well as thorium decay (Table 1). chains were measured using a high-resolution Since the gauging stations on the Meana and gamma-ray spectrometer. To calculate the effective Chacha do not exist anymore and it is unclear how dose rates, the activity concentration results were the discharge measurements were taken, some errors combined with information on the burial, water, and cannot be excluded. organic content history. The resulting dose rates are The data were analyzed using standard statisti- listed in Table 2. For more information on dose rate cal procedures in SPSS 10.0, Microsoft Excel and determination and the OSL method in general, the Libre Office Calc. Visualization was performed with reader is referred to Preusser et al.37 the GIS software packages Saga-GIS 2.1.4, To determine the palaeodose of the samples, the Quantum-GIS 2.1, and Esri ArcGis 10.1. single-aliquot regenerative-dose (SAR) protocol38 was applied to purified sand-sized quartz extracts (90–180 μm fraction). The reliability of the measure- Age Determination ment procedure was checked by a dose recovery During fieldwork, several sediment trenches were experiment; an average dose recovery ratio of investigated to help understand the respective fluvial, 0.99 Æ 0.05 (n = 14) confirmed the suitability of the geomorphological or climatological history. From measurement setup. To obtain a meaningful estimate these sediment profiles, seven charcoal pieces were of the palaeodose received by the samples during bur- collected for radiocarbon dating and six bulk samples ial the palaeodose measurements were repeated on for Optically Stimulated Luminescence (OSL) dating. 18–39 aliquots per sample. The single aliquots palaeo- OSL sampling was conducted on Eolian loess dose distributions of the two younger samples under or sandy loess samples, taken in opaque plastic tubes investigation (Turk-OSL 1 and 2) were significantly

TABLE 1 | The Utilized Hydroclimatic Datasets Station Unit Period Resolution Location Elevation (m) Meana M/C m3/second 1936–1985 Monthly 36.8780 N 60.3875 E 299 Gridded mm 1951–2000 Monthly Cell center: 60.25 E 36.75 N — Hawdan mm 1961–1990 Monthly 37.6515 N 58.4025 E 1480 Kaka mm 1939–1991 Daily 37.3460 N 59.6149 E 293 Mashhad C 1950–2003 Daily 36.3270 N 59.7903 E 927

6of20 ©2017WileyPeriodicals,Inc. Volume4,March/April2017 WIREs Water Modern impacts on piedmont plain overdispersed, i.e., the palaeodose distributions were 2. the hydrological and climatological analysis, more scattered than expected based on the measure- which are discussed with respect to the ﬂuvial ment uncertainties. Given the palaeodose distributions and irrigation historical context. (right-skewed) obtained and the depositional environ- ment of the two younger samples, this scatter is most likely caused by heterogeneous signal resetting (poor Chronostratigraphic Analyses bleaching). To obtain meaningful palaeodose esti- The radiocarbon and luminescence age determina- mates from those scattered palaeodose distributions, tions are presented in their chronological order and we employed the bootstrapped Minimum Age Model discussed in the next paragraph. (bootMAM: Ref 39). The two older samples (Turk- The two oldest OSL ages are the samples ‘Turk OSL 5 and 6) show no indication of heterogeneous OSL 5’ and ‘Turk OSL 6’ at Site 1 (Table 2 and signal resetting. For these two samples the palaeodose Figure 4). They were taken from a small loess lens, was calculated using the Central Age Model (CAM: intercalated in an open gravel quarry. Sample ‘Turk Ref 40). The resulting palaeodoses of the samples were OSL 5’ was taken 160 cm below the surface (of the divided by the dose rates to obtain the burial ages of gravel quarry) and points to a late Pleistocene age of the samples. Palaeodoses and ages are listed in Table 2 29.8 Æ 1.5 ka. Sample ‘Turk OSL 6’ was taken together with an estimate of the validity. 30 cm below the surface (of the gravel quarry) at the Radiocarbon dating was performed by Acceler- top of the gravel and points to a very late Pleisto- ator Mass Spectrometry (AMS) on charcoal pieces at cene/Younger Dryas age of 13.6 Æ 0.9 ka. the Poznan Radiocarbon Laboratory (http://www. From Site 2, three OSL samples were taken radiocarbon.pl/). Calibration was carried out by the from the banks of the Wadi Meana. Sample ‘Turk laboratory using OxCal v4.1.5 Bronk Ram- OSL 1’ was sampled at 135 cm below the modern sey (2010). surface and reveals an age of 3.1 Æ 0.6 ka. Sample ‘Turk OSL 2’ was taken at the same location at RESULTS AND DISCUSSION 70 cm below the modern surface and reveals an age of 2.5 Æ 1.1 ka. Sample ‘LUM 1’ was taken 275 cm The results presented encompass the following below the modern surface and reveals an age of points: 3.4 Æ 0.1 ka. The sample ‘LUM 4’ reveals an early Holocene 1. the chronostratigraphic analysis, which is dis- age of 8.5 Æ 0.2 ka. It was taken in the context of cussed with respect to the broader landscape the archeological excavation at the site of MD (Ref historical context; and 41 in Ref 42).

TABLE 2 | OSL-Dating: (a) Sample Location and Age Determinations, and (b) Laboratory Results (a) Sample Lat. N Lon. E ELV (m a.s.l.) Depth (m) Age (ka) Site Turk OSL 1 36.870491 60.383136 299 1.35 3.1 Æ 0.6 Site 2 Turk OSL 2 36.870491 60.383136 299 0.7 2.5 Æ 1.1 Site 2 Turk OSL 5 36.88974 60.36678 290 1.6 29.8 Æ 1.5 Site 1 Turk OSL 6 36.889975 60.366219 290 0.3 13.6 Æ 0.9 Site 1 LUM4 36.872367 60.3881 290 3.66 8.5 Æ 0.2 MD LUM1 36.872367 60.3881 300 2.75 3.4 Æ 0.1 Site 2 (b) Palaeodose Water Sample NCL Code (Gy) Content (%) LOI (%) Dose Rate (Gy/ka) Systematic Random Validity Comments Turk OSL 1 NCL-7114003 7.5 Æ 1.4 1.15 Æ 0.29 0.95 Æ 0.09 2.46 Æ 0.08 0.1 0.57 Q bootMAM Turk OSL 2 NCL-7114004 6.1 Æ 2.7 1.15 Æ 0.29 0.95 Æ 0.09 2.47 Æ 0.08 0.08 1.09 Q bootMAM Turk OSL 5 NCL-7114005 76.9 Æ 3.1 1.47 Æ 0.37 1.11 Æ 0.11 2.58 Æ 0.08 1.01 1.15 lo CAM Turk OSL 6 NCL-7114006 32.4 Æ 1.8 1.56 Æ 0.39 0.84 Æ 0.08 2.37 Æ 0.08 0.45 0.78 lo CAM bootmam, minimum age model; cam, central age model; lo, likely okay; MD, Monjukli Depe; Q, questionable.

FIGURE 4 | The presented proﬁles and their sedimentological sequence (for pictures, see Appendix C).

The oldest 14C-sample is ‘Turk_9,’ which is began. It is assumed that the dry cold steppes of the calibrated to a Middle Holocene date of 5983–5746 Pleistocene and Younger Dryas were successively BCE. It was taken in the context of the archeological replaced by forest-steppe vegetation with a maximum excavation at MD (the sample and the site descrip- increase of arborescent species.46,47 At a remarkably tion is published in Ref 41 in Ref 42). early time, people started to populate the area, set- Two charcoal samples ‘Turk 4_1’ were taken at tling in the area toward the end of the Neolithic some Site 3 from the infill of a small wadi, at 65 and 8000 years ago. They probably introduced a whole 70 cm below the surface, respectively (see Figure 4). ‘Neolithic package’ of domesticated plants and ani- They reveal statistically identical Mid-Holocene mals, among them sheep, goat, and dog, as well as calendric ages of 5208–4853 BCE and 5207–4855 einkorn and two-row barley, and a largely sedentary BCE. Three samples were taken at Site 4, at the side way of life (Ref 48, p. 33–44 and Ref 2). walls from a 5-m-deep incised wadi near Chacha (see For the broader area of lower Turkmenistan, Figure 4 and Appendix C). The samples ‘Turk 6_1’ climatic changes and oscillations through the Holo- reveal late Holocene to modern calendric ages of cene were probably minor, as indicated by several 424–565 CE at 410 cm depth, 1691–1925 CE at biostratigraphic, geomorphological, and archeologi- 80 cm depth, and 1679–1940 CE at 30 cm cal proxy data.3,45,47,49 In archeological terms, the (Table 3). region was more or less continuously settled from the Neolithic to at least Late Sasanian times (Castro Gessner in Ref 42). Several small Neolithic to early Landscape Historical Context Aeneolithic settlements of the seventh to late fifth Since the Pleistocene the piedmont plain of Turkme- millennia are followed by increasingly complex settle- nistan has experienced times of changing climatic ment systems in the Later Aeneolithic with Ilgynly and hydrological conditions. It was subject to differ- Depe (see Ref 50) as a center, and later, in the late ent dominating landscape formation processes such third and early second millennia BCE, Altyn Depe as as mass movements, Eolian or fluvial processes, and a major city.51 Both of these population concentra- human activity. tions must have had a sustained supply of water con- For the Pleistocene penultimate and last glacial sisting of a canal system that provided these cycle, more arid conditions are assumed for the extensive settlements with drinking water. Judging by Kopet Dag, including increased Eolian activity and their closeness to the Chacha or Meana rivers, probably increased erosional activity, which led to respectively, these canals were likely rather short. the deposition of coarse materials (cf. Refs 43,44). These relatively simple irrigation technologies led to The deposition of this coarser material of Pleis- settlement patterns in which major sites were con- tocene age is attested by the sample OSL Turk 5 inter- strained to locations favored by natural water flow. calated in the quarry gravel formation (Table 2). At Unfortunately, much of the nonurban use of this the end of the Pleistocene/beginning of the Holocene, landscape as well as the intricacies of the settlement a last drop to colder and more arid conditions proba- patterns of the prehistoric periods, including not just bly occurred during the Younger Dryas stage, which small sites but also features such as canals or terra- is represented by the OSL Turk 6 sample in the upper cing, remain concealed under thick layers of loess. part of the gravel (Table 2 and Appendix).45 However, for the Middle Bronze Age (Namazga V With the onset of the Holocene, the climate period), a reconstruction of a highly centralized set- became more humid and a period of relative stability tlement pattern appears likely, with Altyn Depe at

TABLE 3 | Radio Carbon Dating Results Calendric Age Sample Lab Lat. N Long. E ELV (m a.s.l.) Depth (cm) 14C Age (Calibrated 2 Sig) Site Turk 4_1 Poz-51988 36,871505 60,482732 267 65 6090 Æ 40 BP 5208–4853 BCE Site 3 Turk 4_1 Poz-51989 36,871505 60,482732 267 70 6090 Æ 35 BP 5207–4855 BCE Site 3 Turk 6_1_a Poz-51990 36,775146 60,51254 297 410 1560 Æ 30 BP 424–565 CE Site 4 Turk 6_1_b Poz-51993 36,775146 60,51254 297 80 75 Æ 30 BP 1691–1925 CE Site 4 Turk 6_1_c Poz-51994 36,775146 60,51254 297 30 120 Æ 30 BP 1679–1940 CE Site 4 Turk_9 Poz-51995 36,84858 60,418182 290 370 6980 Æ 50 BP 5983–5746 BCE MD

Dating material was charcoal.

Volume4,March/April2017 ©2017WileyPeriodicals,Inc. 9of20 Focus Article wires.wiley.com/water the top of the settlement hierarchy and much smaller interpretation is difficult, because different reasons secondary satellite sites around it. In the Iron Age might cause the deposition of the loess layer, which (late second to first half of the first millennium BCE) here is up to several meters thick. Reasons might the population must have shrunk, only to pick up be: (1) less vegetation or changing wind directions again in the Sasanian period. During that time, an in the potential source areas, leading to intensified unprecedented number of people settled in numerous deflation, (2) more vegetation in the sink area, large sites that were more evenly spread out across leading to more deposition, (3) the fluvial redeposi- the Kopet Dag foothills (Castro Gessner in Ref 42). tion of loess, which is readily available in the lower This is the most likely period for the introduction of Kopet Dag range, and (4) the alteration of vegeta- a qanat irrigation system. The Sasanian system, with tion or fluvial patterns by humans.58 However, the its unconstrained spread of larger sites across the incision of the Wadi Meana through this loess foothill zone, differs quite markedly from the Bronze layer and into its own gravel bed shows the relative Age one, which was dependent on canals leading stability of the valley position (Ref 42, see picture from the rivers. Qanat irrigation frees settlement in Appendix C). locations from dependence on above-ground natural The latest three dates were taken in a branch of water courses. Traces of such a system can still be the Chacha river and also refer to loess depositional found in the area today. processes. The river branch is interpreted as a mod- In fact, the piedmont area is better character- ern gully incision that naturally exposes piedmont ized as an ecosystem that does not primarily depend plain sediments. The dated charcoal remains, which on its local but rather regional hydroclimatic condi- range from the first millennium CE to the present tions, here the adjacent Kopet Dag mountains. In this (samples Turk_6 a,b,c), were taken from a context of sense, it is better compared to the numerous central archeological layers buried under loess including pot- Asian riverine or oasis landscapes that have long tery (picture in Appendix C). At least the earlier date been nuclei of human settlements and that depend matches the archeological evidence of cultural phases predominantly on their respective ‘water towers.’10 dating to Sasanian times. The fact that these cultural Hence changing hydrological conditions in distant layers are enclosed by loess-like material indicates areas appear to be of higher significance concerning that loess and Eolian deposition may have prevailed questions of the regional irrigation history than local since the Middle Holocene. This might resemble the conditions. Possible distant drivers of hydrological modern dust activity reported for Turkmenistan in changes are monsoonal patterns in southeast Asia, general.31 glacier melt in the central Asian highlands (Pamir Shan, Altai, and Tian Shan) and the subsequent change in fluvial patterns, e.g., of the Amu Darya Hydrological and Climatological Analyses River or of the disappeared Usboy River, and finally Several small rivers drain the Kopet Dag Mountains, – the Holocene oscillations of the Aral Sea.52 55 flow eastward and eventually desiccate in the Kara- The only weak indication for a slight change in kum Desert. Two of these, the Meana and Chacha fluvial activity during Mid-Holocene in the area are rivers, are the focus of this analysis. It is assumed the two charcoal samples Turk 4_1 a and b which that their fluvial history and runoff behavior together were taken from the fill material of a small Meana with their climatological setting are representative of tributary. However, the interpretation of a distinct much of the piedmont area concerning its natural change or higher fluvial activity, and radiocarbon water budget. Both rivers show very similar runoff dates of fluvial regimes are problematic.25,56 behavior and their catchment character is compara- During the transition from the Middle to the ble (Table 4). The mean annual discharge of the Late Holocene, the area must have experienced a Meana river for the observed period is 0.013 km3 major change in Eolian depositional processes. This and that of the Chacha 0.011 km3. Both have their is manifest in the two neighboring sections from peak runoff in April. the banks of the Wadi Meana. Here a thick loess By comparing the discharge of both rivers in layer overlies the older river gravel. The OSL dates Figure 5(a) and (b), two features are evident: (1) their from the lowest part approximate the onset of this runoff behavior is very similar and (2) they show a loess deposition between 3.1 Æ 0.6 and distinct reduction in discharge between 1965 3.4 Æ 0.1 ka (samples Turk OSL 1 and LUM and 1985. 1, respectively). Comparative active loess deposi- For (1): The monthly discharge values of the tions are reported from the Iranian plains and Meana and Chacha show a very high correlation might point to more arid conditions.43,57 However, (R2 = 0.98 and Pearson coefficient = 0.99). So to

TABLE 4 | Catchment Character of the Meana and Chacha River Area Mean Discharge from 1936 Mean Discharge from 1965 Reduction Between Catchment (km2) Min_Elv Max_Elv to 1965 (m3/second) to 1985 (m3/second) Discharge Periods (%) Chacha 2219.9 332 2606 0.4 0.16 63 Meana 1702.0 268 2622 0.45 0.18 66

FIGURE 5 | The hydrological year of (a) the Chacha river and (b) the Meana river. reduce the scope of the following analysis, it is suffi- Meana river. The maximum precipitation at this sta- cient to display the Meana river observations (see tion is in April, as is the maximum observed dis- also Appendix A). charge. The weather station of Kaka is assumed to For (2): From 1965 onward, the Meana and represent the climatic conditions on the lower pied- Chacha rivers’ mean annual runoff values perma- mont plain, which shows less agreement with the dis- nently drop below the all-year mean. For the Meana, charge values (R2 = 0.57) (details and data in the the reduction in discharge from 1965 onward is Appendix B). approximately 66% and for the Chacha it totals In a last step, the discharge history with the approximately 63%. In addition, the previously high pronounced drop in discharge was compared to the April value is evened out. climatic datasets over time. Especially the precipita- In a first step, the monthly runoff values were tion was checked for indications of altering condi- compared to monthly precipitation data (Figure 6). tions or climatic change. In fact, none of the The result shows that precipitation data from the examined climatic datasets show distinct changes or Hawdan weather station in the Kopet Dag best fits trends over the respective years, but stay within their the seasonal and monthly discharge (R2 = 0.82), standard deviations over their entire time series despite its location about 200 km away from the (Figure 7 and Appendix B).

FIGURE 6 | Annual discharge of the Meana in comparison to monthly precipitation values of the Kaka and Hawdan weather stations.

3 Discharge (meana) 500 ﬂ Hawdan (highlands) ow patterns. These oscillations also lead to new Kaka (lowlands) river or wadi incisions and branches, which subse- 400 quently lead to active and inactive parts of the allu- 2 300 vial fans. This then implies shifts of agriculturally

/second) 42 3 used areas and/or shifts in settlement locations. The ﬂ 200 controlling mechanisms of these ow patterns are 1 complex and mostly dominated by upstream pro- Discharge (m 100 Annual precipitation (mm) cesses and conditions, i.e., the climatic and hydrological conditions, lithology and tectonic activity, and

0 0 human impact. Moreover, a stream is controlled in 1940 1950 1960 1970 1980 1990 its lower reaches by its base level, its valley morphol- FIGURE 7 | Annual time series of the Meana river’s discharge ogy, the prevailing vegetation (density), flood fre- 60 and precipitation at Kaka and Hawdan. quency, and again human agency. The archeological record implies that human However, the datasets used differ in their reso- agency manifest in irrigation measures must have lution, period covered, and distance to the area of played a major role. These measures might have ran- interest, and must be interpreted with caution. ged from simple floodwater techniques with small canals or ditches and leveling, to larger scale techniques of dam building or water storage Fluvial Landscape History (cf. Refs 59,61). The modern diversion of water from the Karakum In addition to the direct utilization of surface or canal broke up traditional irrigation measures and river discharge water, groundwater must have played obscures much of those former usages, making their a much greater role in the past than at present. detailed geomorphological and hydrological recon- Groundwater tapping conduits or tunnels, so-called struction difficult.11 qanats, which are well-known from the Iranian pla- Especially for the Early and Mid-Holocene, it teau, are also widespread in the Turkmen piedmont cannot be excluded that there were times of higher zone.62,63 In the investigated area, several qanats rainfall availability in the lowlands that permitted totaling about approximately 20 km in length are rain-fed agriculture. However, the importance of the still visible on aerial imagery (indicated in Figure 2). Kopet Dag rivers for the piedmont plain and its They directly account for a former groundwater millennia-long irrigation schemes appears obvious.59 usage, most probably for irrigation. Since they are This is strengthened by the fact that the continuous out of use and dry—as was checked at several water and sediment supply through the alluvial fans shafts—together with dried-out water wells in the onto the floodplains provide the most preferable area, it is assumed that the groundwater level locations for irrigation endeavors.17 dropped at some point in time. This might well be However, the courses of the Kopet Dag rivers connected with the overall observed drop in runoff are not stable and show anastomosing and changing discharge in the second half of the 20th century. The

12 of 20 ©2017WileyPeriodicals,Inc. Volume4,March/April2017 WIREs Water Modern impacts on piedmont plain large drop observed by comparing discharge volumes the stream bed is a reasonable maximum for agricul- for the years 1936–1965 and 1965–1985, which is tural field extents. Assuming this simple 1 km buffer, also confirmed by the Russian maps from the 1950s an area of 87 km2 would account for the irrigable where both Meana and Chacha rivers are labeled as area and the amount of water that is held back in the perennial, is remarkable. upper catchment. Comparing the two respective time What follows are attempts to evaluate this frames, this would total a mean annual retention of decrease, since there appear to be no straightforward 12 million cubic meters of water per square kilometer explanation(s) for it. As shown above neither the (or a water surplus of ~3500 L/day and hectare). rainfall data nor the temperature at Mashhad, which Since this study is unfortunately restricted to was checked for possible trends in evaporation or the Turkmen side of the valleys, this very high seasonal changes in snowfall, show any clear trends amount of reduced water flow cannot be investigated through their respective time series which could in appropriate detail and remains an unsolved prob- account for the observed runoff decrease. This lem for future studies. Contributing explanations implies that climatic changes should be excluded as might be: (1) an expansion of river water abstraction the dominant or prime cause. Other physical pro- and intensified irrigation with surface waters, or (2) a cesses such as tectonic activity and earthquakes or higher outtake of groundwater from wells. Both changes in lithology cannot be excluded, but such might lead to either higher infiltration capacities or evidence is neither published nor does it appear very soil water contents due to different surface properties likely to cause such continuous decrease. that might increase evaporation rates, e.g., from open This leaves human water diversion or a change fields or water storages (cf. Ref 66). in water management strategies as the most likely It should be considered as a hydrological curi- reasons. The most obvious sources for such runoff osity that the drop in natural water supply happened decreases are the constructions of river dams.64,65 exactly during the time in which the piedmont plain However, no dams have been built in the upper of Turkmenistan became detached from dependency catchment of the two rivers in question during the on this supply. Since the mid-1960s water could be time frame considered, and a cursory personal com- taken from the Karakum canal. The construction munication with local water authorities did not time of the canal matches the time of continuous reveal any such information (a list of Iranian dams decrease of runoff from the upper catchments, so that from the FAO was checked: see reference AQUA- the latter would not have sustained the millennia- STAT). The only dam-like construction visible on long irrigation schemes on the piedmont any longer. recent aerial or satellite imagery is in the upper catchment of the Meana river (at 36.798988N, 60.078896 E). It was probably built around 1988–1994 and hence later than the time covered CONCLUSIONS here with its observed decrease in discharge. This dam may well, however, contribute to the observed The main outcome of this study is that an irrigation dryness of the recent years in the lower Meana valley system that most probably worked for many millen- (personal communication with local persons). nia broke down in the middle of the 20th century. It Hence the drop in discharges must be explained appears that it was not climate, but rather changing by current land- and water management in the upper water management and land use in the upper Iranian Iranian catchments. parts of the investigated catchments that led to the A simple GIS analysis of the Meana catchment remarkable drop in discharge in the lower reaches of delineates the upper basin into one major and two the Chacha and Meana rivers. This drop in discharge minor tributaries, with total flow lengths of and the coincident availability of the waters from the about 87 km. The streams appear to represent typi- Karakum canal most probably led to a loss of local cal V-shaped mountain valleys with some wider areas knowledge concerning historical irrigation and water (pools) that can be used for cultivation. As indicated harvesting techniques that were well adapted to local by aerial and satellite imagery, a 1-km strip around pre-20th century conditions.

ACKNOWLEDGMENTS The study is part of the Cluster of Excellence Exc264 TOPOI ‘The Formation and Transformation of Space and Knowledge in Ancient Civilizations’ (http://www.topoi.org/).

REFERENCES 1. Dolukhanov PM. The ecological prerequisites for early 15. Pollock S, Bernbeck R, Benecke N, Castro Gessner G, farming in southern Turkmenia. In: Kohl P, ed. The Daszkiewicz M, Eger J, Keßeler A, Miller N, Pope M, Bronze Age Civilization of Central Asia: Recent Soviet Ryan P, et al. Excavations at Monjukli Depe, Meana- Discoveries. Armonk, NY: M.E. Sharpe; 1981, Chacha Region, Turkmenistan, 2010. AMIT 2011, 359–385. 44:1–69. 2. Harris DR. Origins of Agriculture in Western Central 16. Francfort H-P, Lecomte O. Irrigation et société en Asie Asia: An Environmental–Archaeological Study. Phila- centrale des origines à l’époque achéménide. Ann Hist delphia: University of Pennsylvania Museum of Sci Soc 2002, 57:625–663. Archaeology and Anthropology; 2010. 17. Lewis RA. Early Irrigation in West Turkestan. Ann 3. Lisitsina GN. The history of irrigation agriculture in Assoc Am Geogr 1966, 56:467–491. doi:10.2307/ southern Turkmenia. In: Kohl P, ed. The Bronze Age 2561780. Civilization of Central Asia: Recent Soviet Discoveries. Armonk, NY: M.E. Sharpe; 1981, 350–358. 18. Lisitsina GN. The earliest irrigation in Turkmenia. Antiquity 1969, 43:279–288. doi:10.1017/ 4. Adams RMC. Heartland of Cities. Chicago, IL: Uni- S0003598X00040710. versity of Chicago Press; 1981. 5. Wilkinson TJ. Archaeological Landscapes of the Near 19. Christian D. Silk roads or steppe roads? The silk roads – East. Tucson, AZ: University of Arizona Press; 2003. in world history. J World Hist 2000, 11:1 26. — 6. Barker G, Gilbertson D. Living at the margin: themes 20. Stone R. A new great lake or dead sea? Science 2008, – in the archaeology of drylands. In: Barker G, 320:1002 1005. doi:10.1126/science.320.5879.1002. Gilbertson DD, eds. The Archaeology of Drylands: 21. O’Hara SL. Irrigation and water management in Turk- Living at the Margin, One World Archaeology. menistan: past systems, present problems and future London and New York: Routledge; 2000. scenarios. Eur-Asia Stud 1999, 51:21–41. 7. Beckers B, Berking J, Schütt B. Ancient water harvest- doi:10.1080/09668139999100. ing methods in the drylands of the Mediterranean and 22. Gutiérrez M, ed. Chapter 15—the arid region pied- – Western Asia. ETopoi J Ancient Stud 2013, 2:145 164. monts: glacis and alluvial fans. In: Developments in 8. Bruins HJ. Ancient desert agriculture in the Negev and Earth Surface Processes. Netherlands: Elsevier; 2005, climate-zone boundary changes during average, wet 343–381. and drought years. J Arid Environ 2012, 86:28–42. 23. Babaev AG. Landscapes of Turkmenistan. In: Fet V, doi:10.1016/j.jaridenv.2012.01.015. Atamuradov KI, eds. Biogeography and Ecology of 9. Mueller EN, Wainwright J, Parsons AJ, Turnbull L, Turkmenistan. Monographiae Biologicae. Netherlands: eds. Patterns of Land Degradation in Drylands. Dor- Springer; 1994, 5–22. drecht: Springer; 2014. 24. Hollingsworth J, Jackson J, Walker R, Reza 10. Karthe D, Chalov S, Borchardt D. Water resources Gheitanchi M, Javad Bolourchi M. Strike-slip faulting, and their management in central Asia in the early rotation, and along-strike elongation in the Kopeh ﬁ twenty rst century: status, challenges and future pro- Dagh mountains, NE Iran. Geophys J Int 2006, – spects. Environ Earth Sci 2014, 73:487 499. 166:1161–1177. doi:10.1111/j.1365-246X.2006. doi:10.1007/s12665-014-3789-1. 02983.x. 11. Nesbitt M, O’Hara S. Irrigation agriculture in Central 25. Marcolongo B, Mozzi P. Holocene geomorphological Asia: a long-term perspective from Turkmenistan. In: evolution of the eastern Kopet Dag piedmont plain: a Graeme Barker, David Gilbertson, ed. The Archaeol- preliminary geoarchaeological perspective. Archaeol ogy of Drylands Living at the Margin. London: Rou- News 2001, 7:38–40. tledge; 2000, 103–124. нжуклы Д п — н г л йн п 12. Hiebert F. The Kopet dag sequence of early villages in 26. Berdiev OK. Mo e e M o oc o oe oce- л ни н лит и нн г н лит в южн м у км Central Asia. Paléorient 2002, 28:25–41. e e eo a pa e oeeo a o T p e- ниcтaнe [Monzhukli-depe:Mnogosloinoe poselenie 13. Kohl P. Central Asia: Paleolithic Beginnings to the neolita i rannego eneo]. Кapaкумcкиe Дpeвнocти [Kar- Iron Age. Editions Recherche sur les Civilisations 14. akumskie drevnosti] 1972, 4:11–34. Paris; 1984. 27. Bernbeck R, Pollock S. Scalar differences: temporal 14. Harris DR, Asouti E, Bogaard A, Charles M, rhythms and spatial patterns at Monjukli Depe, south- Conolly J, Coolidge J, Dobney K, Gosden C, ern Turkmenistan. Antiquity 2016, 90:64–80. Heathcote J, Jaques D, et al. Origins of Agriculture in Western Central Asia: An Environmental- 28. Pollock S, Bernbeck R, Ögüt B, et al. The second sea- Archaeological Study. Philadelphia: University of Penn- son of excavations at Monjukli Depe, Turkmenistan, sylvania Press; 2010. 2011. AMIT 2013, 44:51–98.

29. Masson VM. Altyn-Depe [Translated by Henry at Monjukli Depe: a regional perspective. AMIT, Berlin: N. Michael]. Philadelphia, PA: University of Pennsylva- In press. nia Press; 1988. 43. Karimi A, Frechen M, Khademi H, Kehl M, Jalalian A. 30. Hiebert FT. Origins of the Bronze Age Oasis Civiliza- Chronostratigraphy of loess deposits in northeast Iran. tion of Central Asia. Cambridge, MA: Harvard Uni- Quat Int 2011, 234:124–132. doi:10.1016/j. versity Press; 1994. quaint.2009.08.002. 31. Orlovsky L. Dust storms in Turkmenistan. J Arid 44. Pierce KL, Scott WE. Pleistoce episodes of alluvial- Environ 2005, 60:83–97. doi:10.1016/j.jaridenv.2004. gravel deposition, southeastern Idaho. In: Bonnichsen 02.008. B, Beckenridge RM, eds. Cenozoic Geology of Idaho: Idaho Bureau of Mines and Geology Bulletin 26. 32. Kharin NG. Desertification of the arid lands of Turk- 1982, 685–702. menistan. In: Fet V, Atamuradov KI, eds. Biogeogra- — phy and Ecology of Turkmenistan. Monographiae 45. Kehl M. Quaternary climate change in Iran the state – Biologicae 72. Dordrecht: Springer; 1994, 65–76. of knowledge. Erdkunde 2009, 63:1 17. doi:10.3112/ Available at: http://link.springer.com/chapter/10.1007/ erdkunde.2009.01.01. 978-94-011-1116-4_5. 46. Leroy SAG, Kakroodi AA, Kroonenberg S, 33. Jin L, Chen F, Morrill C, Otto-Bliesner BL, Lahijani HK, Alimohammadian H, Nigarov A. Holo- Rosenbloom N. Causes of early holocene desertifica- cene vegetation history and sea level changes in the SE tion in arid Central Asia. Clim Dyn 2012, 38:1577– corner of the Caspian Sea: relevance to SW Asia cli- – 1591. doi:10.1007/s00382-011-1086-1. mate. Quat Sci Rev 2013, 70:28 47. doi:10.1016/j. quascirev.2013.03.004. 34. Rondelli B, Stride S, García-Granero JJ. Soviet military maps and archaeological survey in the Samarkand 47. Lioubimtseva E, Cole R. Uncertainties of climate change region. J Cult Herit 2013, 14:270–276. doi:10.1016/j. in arid environments of Central Asia. Rev Fish Sci 2006, – culher.2012.06.006. 14:29 49. doi:10.1080/10641260500340603. 48. Masson VM, Sarianidi VI. Central Asia: Turkmenia 35. Beck C, Grieser J, Rudolf B. A New Monthly Precipita- Before the Achaemenids [Translated and Edited by tion Climatology for the Global Land Areas for the R. Tringham]. London: Thames & Hudson; 1972. Period 1951 to 2000. DWD; 2005. 49. Tarasov PE, Webb T III, Andreev AA, Afanas’eva NB, 36. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Berezina NA, Bezusko LG, Blyakharchuk TA, et al. Jarvis A. Very high resolution interpolated climate sur- Present-day and mid-holocene biomes reconstructed faces for global land areas. Int J Climatol 2005, from pollen and plant macrofossil data from the 25:1965–1978. doi:10.1002/joc.1276. former Soviet Union and Mongolia. J Biogeogr 1998, 37. Preusser F, Degering D, Markus F, Hilgers A, Kadereit 25:1029–1053. doi:10.1046/j.1365-2699.1998.00236.x. A, Klasen N, Krbetschek M, Richter D, Spencer JQG. 50. Solovyeva N. Chalcolithic Anthropomorphic Figurines Luminescence dating: basics, methods and applica- from Ilgynly-depe, Southern Turkmenistan: Classifica- tions. Recent Prog Quatern Dating Methods 2008, 57 tion, Analysis and Catalogue (BAR S1336). Oxford: – (1/2):95 149. Archaeopress; 2005. 38. Murray AS, Wintle AG. The single aliquot regenerative 51. Masson VM. The urban revolution in south Turkme- dose protocol: potential for improvements in reliabil- nia. Antiquity 1968, 142:178–187. ity. Radiat Meas 2003, 37:377–381. doi:10.1016/ S1350-4487(03)00053-2. 52. Andrianov BV, Kes AS. The Development of Hydrog- raphy Net of Irrigation in the Central Asian Plains: 39. Cunningham AC, Wallinga J. Realizing the potential Problems of Nature Reform in Central Asia fl of uvial archives using robust OSL chronologies. (In Russian). 1967. Quat Geochronol 2012, 12:98–106. doi:10.1016/j. 53. Boomer I, Aladin N, Plotnikov I, Whatley R. The quageo.2012.05.007. palaeolimnology of the Aral Sea: a review. Quat Sci 40. Galbraith RF, Roberts RG, Laslett GM, Yoshida H, Rev 2000, 19:1259–1278. doi:10.1016/S0277-3791 Olley JM. Optical dating of single and multiple grains (00)00002-0. of quartz from jinmium rock shelter, Northern 54. Boroffka N, Oberhänsli H, Sorrel P, Demory F, Australia: part I, experimental design and statistical Reinhardt C, Wünnemann B, Alimov K, Baratov S, models. Archaeometry 1999, 41:339–364. Rakhimov K, Saparov N, et al. Archaeology and climate: 41. Berking J, Beckers B. Landscape setting and fluvial his- settlement and lake-level changes at the Aral Sea. Geoarch- tory of Monjukli Depe. In: Pollock S, Bernbeck R, aeology 2006, 21:721–734. doi:10.1002/gea.20135. Ögüt B, et al., eds. Archaeological Work at Monjukli 55. Létolle R, Micklin P, Aladin N, Plotnikov I. Uzboy Depe: A Regional Perspective. Berlin: AMIT; In press. and the Aral regressions: a hydrological approach. 42. Pollock S, Bernbeck R, Beckers B, Benecke N, Berking J, Quat Int 2007, 173–174:125–136. doi:10.1016/j. Castro Gessner G, Eger J, Ögüt B. Archaeological work quaint.2007.03.003.

56. Hoffmann T, Lang A, Dikau R. Holocene river activ- 61. Fleskens L, Ataev A, Mamedov B, Spaan WP. Desert ity: analysing 14C-dated fluvial and colluvial sediments water harvesting from Takyr surfaces: assessing the from Germany. Quat Sci Rev 2008, 27:2031–2040. potential of traditional and experimental technologies doi:10.1016/j.quascirev.2008.06.014. in the Karakum. Land Degrad Dev 2007, 18:17–39. doi:10.1002/ldr.759. 57. Schmidt A, Quigley M, Fattahi M, Azizi G, fi Maghsoudi M, Fazeli H. Holocene settlement shifts 62. Langbein J. Deserti kationsbekämpfung und Ressour- and palaeoenvironments on the Central Iranian Pla- cenmanagement im Kopet-Dag, Turkmenistan: teau: investigating linked systems. Holocene 2011, Akteure und ihre Handlungsspielräume. Doctoral The- 21:583–595. doi:10.1177/0959683610385961. sis, Universität zu Köln, 2006. 63. Ovezberdyyeva A. Sustainable water management in 58. Tsoar H, Pye K. Dust transport and the question of Turkmenistan: challenges and solutions. Master The- desert loess formation. Sedimentology 1987, sis, University of Greifswald, 2009, 1–102. 34:139–153. doi:10.1111/j.1365-3091.1987.tb00566.x. 64. AlMaarofi S, Douabul A, Al-Saad H. Mesopotamian 59. Harris DR, Masson VM, Berezkin YE, Charles MP, marshlands: salinization problem. J Environ Prot Gosden C, Hillman GC, Kasparov AK, 2012, 03:1295–1301. doi:10.4236/jep.2012.310147. Korobkova GF, Kurbansakhatov K, Legge AJ, 65. Yang Z, Zhou Y, Wenninger J, Uhlenbrook S. The et al. Investigating early agriculture in Central Asia: causes of flow regime shifts in the semi-arid Hailiutu new research at Jeitun, Turkmenistan. Antiquity 1993, River, Northwest China. Hydrol Earth Syst Sci 2012, 67:324–338. doi:10.1017/S0003598X00045385. 16:87–103. doi:10.5194/hess-16-87-2012. 60. Schumm SA. Geomorphic thresholds and complex 66. Gallart F, Llorens P. Catchment management under response of drainage systems. In: Morisawa W, environmental change: impact of land cover change on ed. Fluvial Geomorphology. Binghamton, NY: Publica- water resources. Water Int 2003, 28:334–340. tions in Geomorphology; 1973, 299–310. doi:10.1080/02508060308691707.

APPENDIX A. DISCHARGE DATA

TABLE A1 Seasonal Discharge Data (m3/s)

Station, Time Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Meana, 1936–1965 0.4 0.4 0.4 0.5 0.5 0.7 1.6 0.7 0.4 0.3 0.3 0.4 Meana, 1965–1985 0.1 0.2 0.2 0.2 0.2 0.2 0.3 0.2 0.1 0.1 0.1 0.1 Chacha, 1936–1965 0.3 0.3 0.3 0.4 0.4 0.7 1.2 0.6 0.3 0.3 0.2 0.3 Chacha, 1965–1985 0.1 0.2 0.2 0.2 0.2 0.2 0.3 0.2 0.1 0.1 0.0 0.1

TABLE A2 Linear Regression and Person Correlation Coefficient of the Meana and Chacha River Discharge (per Month)

RR2 Standard Error Correlation (Pearson) Sig. (Pearson) 1 0.991 0.982 0.035 0.991 .000

APPENDIX B. METEOROLOGICAL DATA

TABLE B1 Monthly Data of the Three Precipitation Datasets (mm/ month) (Note That the Gridded Precipitation Datasets Was for the Sake of Clarity Deleted From Figure 6)

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Annual Gridded data 31.8 33.6 50.3 41.8 21.0 2.5 0.4 0.4 1.3 7.5 15.1 24.4 230.0 Kaka 27.6 25.5 45.8 39.5 23.4 3.3 0.3 0.7 1.8 10.4 13.6 18.6 210.3 Hawdan 29.2 32.6 56.7 69.5 52.9 13.2 7.9 3.8 7.9 18.6 21.2 26.5 340.0

TABLE B2 Linear Regression of the Monthly Precipitation Values With the Monthly Meana River Discharge

Meana Mean Monthly (m3/s) R2 With Kaka 0.57 With Hawdan 0.82 With Grid 0.49

TABLE B3 Descriptive Statistics and Linear Regression of Long Term Trend of Each Datasets

Discharge Precipitation Precipitation Precipitation Min. Temperature (Meana) (Hawdan) (Kaka) (Gridded) (Mashhad) N (years) 53 30 44 46 34 Mean 0.4 340.1 210.3 230.2 8.1 Minimum 0.0 180.0 120.0 143.0 4.9 Maximum 2.8 474.0 311.7 340.0 12.1 Standard deviation 0.4 76.7 50.7 51.3 1.5 Linear trend with 0.39 0.05 0.13 0.0 0.14 time (R2)

APPENDIX C. PHOTOS OF THE (SAMPLING) SITES

FIGURE C1 Compilation of the four sampling sites (black stars indicate sampling points).

TABLE C1 Sampling Sites and Chronological Ages (Ages Are “ka cal. BP” for Radiocarbon Values Respectively)

Sample Proﬁle Site Fazies Type Age (ka) Phase OSL 5 Gravel_Pit Site 1 Gravel OSL 29.8 Æ 1.5 Pleistocene OSL 6 Gravel_Pit Site 1 Gravel OSL 13.6 Æ 0.9 Pleistocene/younger Dryas Turk 4_1 Dry_Wadi Site 3 Silt 14C 6.98 Æ 0.18 Middle Holocene Turk 4_1 Dry_Wadi Site 3 Silt 14C 6.98 Æ 0.18 Middle Holocene LUM1 Wadi_Meana Site 2 Silt OSL 3.4 Æ 0.1 Middle Holocene OSL 1 Wadi_Meana Site 2 Silt OSL 3.1 Æ 0.6 Middle Holocene OSL 2 Wadi_Meana Site 2 Sand OSL 2.5 Æ 1.1 Middle Holocene Turk 6_1 Chacha_Gully Site 4 Silt 14C 1.456 Æ 0.07 Late Holocene or modern Turk 6_1 Chacha_Gully Site 4 Silt 14C 0.142 Æ 0.12 Late Holocene or modern Turk 6_1 Chacha_Gully Site 4 Silt 14C 0.141 Æ 0.13 Late Holocene or modern

FIGURE C2 The Wadi Meana (coordinates: 36.8726N; 60.3904E).

FIGURE C3 The piedmont plain, the loess hills, and the Kopet Dag mountains (coordinates: 36.8454N; 60.4147E).

FIGURE C4 The Gully near Chacha and sample area “Turk_6” (coordinates: 36.7755N; 60.5150E).